Characterization of the tension softening behavior of UHPC: Fid-Bau Portal

Characterization of the tension softening behavior of UHPC

Saqif, M.A. / El-Tawil, Sherif

Highlights • Fiber type, length, and volume fraction influence UHPC's tension softening behavior. • DTTs are used to formulate an empirical relationship for the tension softening region. • The proposed tension-softening curves are effectively validated using 4PBT specimens. • Response of surface of softening parameters can be utilized in finite element (FE) modeling without resorting to performing DTTs.

Abstract Fracture simulation of concrete structures, which entails generating rational and accurate crack growth through the material, requires the tension softening curve as a constitutive property. Unlike regular concrete, the tension softening behavior of ultra-high performance concrete (UHPC) is highly pronounced and extends to large deformation levels. However, data about this property and, especially, how it is affected by key variables (i.e., fiber dosage and type of fiber used) is lacking. In this paper, a set of direct tension tests (DTT) are conducted to formulate an empirical relationship for UHPC in the tension softening range. The experimental variables are fiber type (two kinds of straight fiber and hooked), fiber length (lf = 13, 19 or 30 mm), and fiber volume fraction (Vf = 1.0 %, 1.5 %, and 2.0 %). The energy absorption capacity of UHPC is found to be different for different type of fibers, due to a difference in strain hardening as well as the pronounced softening phase. A dimensionless tension-softening curve for UHPC that is a function of fiber type, length and volume fraction is proposed as a generalized softening expression (expressed using two parameter a and b) and then implemented into a hybrid rotating/fixed crack smeared crack model. Four-point bending test (4PBT) specimens that correspond to the DTT specimens are made and tested to provide validation data. These validations confirm the model's adeptness at accurately emulating all facets of the 4PBT specimen's response. Following the validation, a response surface for parameters 'a' and 'b' is developed, enabling the generation of the softening curve across the entire spectrum of lf, Vf, and fiber type, without having to monitor the entire range of the softening response.